Improving the scaling normalization for high-density oligonucleotide GeneChip expression microarrays.

摘要: Normalization is an important step for microarray data analysis to minimize biological and technical variations. Choosing a suitable approach can be critical. The default method in GeneChip expression uses constant factor, the scaling factor (SF), every gene on array. SF obtained from trimmed average signal of array after excluding 2% probe sets with highest lowest values. Among 76 U34A experiments, total signals each showed 25.8% variations terms coefficient variation, although all microarrays were hybridized same amount biotin-labeled cRNA. that normally excluded calculation accounted 34% 54% (40.7% ± 4.4%, mean sd). In comparison normalization factors median or log transformed signal, greatest variation. least Eliminating 40% during failed show any benefit. performed best. Thus, it suggested use logarithm normalization, rather than arithmetic microarrays.